Understanding and Mitigating Memorization in Generative Models via Sharpness of Probability Landscapes

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Jeon, Dongjae, Kim, Dueun, No, Albert
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915450903855104
author Jeon, Dongjae
Kim, Dueun
No, Albert
author_facet Jeon, Dongjae
Kim, Dueun
No, Albert
contents In this paper, we introduce a geometric framework to analyze memorization in diffusion models through the sharpness of the log probability density. We mathematically justify a previously proposed score-difference-based memorization metric by demonstrating its effectiveness in quantifying sharpness. Additionally, we propose a novel memorization metric that captures sharpness at the initial stage of image generation in latent diffusion models, offering early insights into potential memorization. Leveraging this metric, we develop a mitigation strategy that optimizes the initial noise of the generation process using a sharpness-aware regularization term. The code is publicly available at https://github.com/Dongjae0324/sharpness_memorization_diffusion.
format Preprint
id arxiv_https___arxiv_org_abs_2412_04140
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Understanding and Mitigating Memorization in Generative Models via Sharpness of Probability Landscapes
Jeon, Dongjae
Kim, Dueun
No, Albert
Machine Learning
Artificial Intelligence
In this paper, we introduce a geometric framework to analyze memorization in diffusion models through the sharpness of the log probability density. We mathematically justify a previously proposed score-difference-based memorization metric by demonstrating its effectiveness in quantifying sharpness. Additionally, we propose a novel memorization metric that captures sharpness at the initial stage of image generation in latent diffusion models, offering early insights into potential memorization. Leveraging this metric, we develop a mitigation strategy that optimizes the initial noise of the generation process using a sharpness-aware regularization term. The code is publicly available at https://github.com/Dongjae0324/sharpness_memorization_diffusion.
title Understanding and Mitigating Memorization in Generative Models via Sharpness of Probability Landscapes
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2412.04140